Standard Forms of SOCP
Second-order cone programs: standard formsInequality formA second-order cone program (or SOCP, for short) is an optimization problem of the form where 's are given matrices, , vectors, and 's scalars. The problem is convex, since the constraint functions of the corresponding standard form are. Examples: Conic formWe can put the above problem in the so-called ‘‘conic’’ format Since the cones are convex, and the mappings are affine, the feasible set is convex. Rotated second-order cone constraintsSince the rotated second-order cone can be expressed as some linear transformation of an ordinary second-order cone, we can include rotated second-order cone constraints, as well as ordinary linear inequalities or equalities, in the formulation. This allows to formulate LPs and QPs as special cases of SOCP. Examples:
Quadratically constrained quadratic programmingDefinitionA quadratically constrained quadratic programming (QCQP for short) is a problem of the form where , . This condition ensures that the problem is convex. QCQPs contain LPs and QPs as special case. Examples: SOCP formulationWe can formulate QCQPs as SOCPs, by introducing new variables and affine equality constraints. (Proof). |